CDA6530:%20Performance%20Models%20of%20Computers%20and%20Networks%20Project%203%20Q - PowerPoint PPT Presentation

About This Presentation
Title:

CDA6530:%20Performance%20Models%20of%20Computers%20and%20Networks%20Project%203%20Q

Description:

Title: PowerPoint Presentation Subject: FSEC template Author: GrossmanN Last modified by: COVE User Created Date: 3/12/2001 8:28:54 PM Document presentation format – PowerPoint PPT presentation

Number of Views:59
Avg rating:3.0/5.0
Slides: 10
Provided by: GrossmanN
Learn more at: http://www.cs.ucf.edu
Category:

less

Transcript and Presenter's Notes

Title: CDA6530:%20Performance%20Models%20of%20Computers%20and%20Networks%20Project%203%20Q


1
CDA6530 Performance Models of Computers and
NetworksProject 3 QA
TexPoint fonts used in EMF. Read the TexPoint
manual before you delete this box. AAAAAAAAAAAA
2
  • m 300 n 300
  • Structure definition in Matlab
  • Nodestruct('status', zeros(m,n), 'infectTime',
    zeros(m,n), 'InfectOtherTime', zeros(m,n))
  • Or you can define
  • NodeState zeros(m,n)
  • NodeInfectTime zeros(m,n)
  • NodeInfectOtherTime zeros(m,n)

3
Multiple Way to Remember Infection Traffic
  • Define a queue variable to remember
  • Generated infection traffic source (or
    destination)
  • Active time (when the source passes the traffic
    to all its neighbors, or when a vulnerble node
    receives the traffic)
  • Quite complicated since the event queue is very
    dynamic

4
Multiple Way to Remember Infection Traffic (my
approach)
  • Use each infected node variable to remember when
    its outgoing infection traffic reach its
    neighbors
  • NodeInfectOtherTime(i,j) save the time for
    infection traffic reaching the node(i,j)s
    neighbors

5
  • How to determine neighboring nodes?
  • You dont need the code to remember the topology
    since it is so regular
  • Node(a,b)s 4 neighbors
  • Upper node (a-1,b), Down node (a1, b)
  • Left node (a,b-1), right node (a, b1)
  • Make sure you check if any of the above 4 nodes
    are non-exist
  • For S2, you also need to check if the node is one
    of those 10 shortcut nodes
  • If yes, considering the 5th neighboring node

6
  • How to decide the simulation end time?
  • At current discrete time k, check all nodes
  • If all (NodeInfectOtherTime(.,.) lt k), then stop
  • It means there does not exist any future
    infection traffic anymore

7
Infection Activity from sending node angle
  • At current time t
  • If the NodeState(j,k) infected
  • If the NodeInfectOtherTime(j,k) t,
  • the node infection traffic reach its neighbors
    now!
  • Check all its neighbors to see if any neighbor is
    infected now, if the neighbor node(a, b) is
    infected now
  • NodeState(a,b) infected
  • NodeInfectTime(a,b) t
  • generate Poisson distr. delay time x
  • NodeInfectOtherTime(a,b) x t 1
  • If NodeState(j,k) vulnerable
  • Do nothing

8
Two Actions for Every Infected Node
  • Act 1 when node(a,b) becomes infected at current
    discrete time t
  • Change its status NodeState(a,b) INFECTED
  • Assign
  • NodeInfectTime(a,b) t
  • x PoissonGenerator()
  • NodeInfectOtherTime(a,b) t x 1
  • Act 2 when an infected node delivers infection
    traffic to its neighbors
  • When? if (NodeState(a,b) INFECTED
  • t
    NodeInfectOtherTime(a,b) )
  • Check each of its neighbor
  • If neighbor node(c,d) is vulnerable and will be
    infected?
  • Run Act 1 for the node(c,d)

9
  • startTime cputime
  • Simulation.
  • simulateCPUTime cputime startTime
Write a Comment
User Comments (0)
About PowerShow.com